Mining Plants Features for Disease Detection Tensor Flow: A Boon to Agriculture

Abstract

The abstract should summarize the contents of the paper and should contain at least 70 and at most 150 words. It’s more difficult, complicated, time consuming and expensive to restore the health of a plant that to prevent illness. Tensorflow and Keras are considered to be the most important and useful libraries for a number of machine learning tasks. In this research paper we have used these two-machine learning techniques for the image mining of diseased plants. These are user friendly and efficient in nature. Lately, Handling of diseases using the traditional methods are laborious, time consuming as well as expensive. Since Agriculture is the primary occupation of 42.38% of Indians, managing of different types of plants and crops becomes very difficult due to the different diseases which arise in them. An estimate of 15–20% of the crops is destroyed in India every year due to pests, diseases and weeds. It is very important to develop an easy and convenient way of identifying these diseases so that it can prevent the loss of crops. This problem can be solved with machine learning technologies. This research paper introduces a novel technique to identify a plant’s heath status with the help of image classification and deep learning algorithms. The dataset is extracted as New Plant Disease data set containing healthy and unhealthy data of plant images. This study evaluates the images of plants taken as input and segregate different species by classifying them into healthy and unhealthy plants. The study uses the mathematical functions to identify dimensions and performs quantitative analysis of the images. The results are observed by evaluating on different metrics with training and validations accuracy as well as loss through visualizations to produce the final outcome as diseased or healthy plants. Train loss: “−0.204”, Train accuracy: “0.03”, Validation loss: “−0.112”, Validation accuracy: “0.0312”.